From: Paul on
Thomas Andersson wrote:
> Paul wrote:
>>> Ok, managed to track his mem down. It's Apacer unbuffered CL3 PC3200
>>> ram and makers page says it uses 32/62MX8 chips so that kills that
>>> specualtion. Mobo is on the list that doesn't support the 64bit ones
>>> as you suggested though so thought was correct.
>>> Hmm, Acers site list only ECC dims for the board though (is that
>>> registered) ? There's nothing about ECC in his current chips specc.
>> The manual says that ECC or non-ECC can be used, but they cannot
>> be mixed. (That is the usual advice regarding the two different
>> types.)
>> Perhaps you could give an exact part number, or a link to the
>> Apacer product page, so I could read the specs.
>> It could be that the RAM is bad, but if you test it one stick at
>> a time, odds are that one of the sticks is still working.
> Whatever it is it's very odd. I asked him to give me all info on the sticks
> he had and he sent.
> On label:
> P/N:77.G1136.23G
> S/N:230816120342
> 1GB UNB PC3200 CL3
> When running apacers ram-finder tool all listed memory is ECC models for
> some reason. Mixed memory types wasn't a problem as he only ran either at
> any time. He tried running both modules one at the time and neither case was
> any improvement.
> Best Wishes
> Thomas

I can find the modules on Internet sale sites, but only in a very restricted
geographical region. I tried searching the Apacer site, and didn't find the
exact number. It is possible I should only be searching for 77.G1136 ,
but I'm still not getting a match. Apacer lists very generic specs,
and didn't give any part numbers, when I looked in their product
section. In any case, looking at all the pictures of modules I could
find, they appear to be non-ECC, with eight chips per side.

What I've run into before, is some module manufacturers use SPD EEPROMs
with invalid data contents. In one case, the capacity listed in the
EEPROM was wrong. For some of them, they just solder any old EEPROM
they have in the bucket. It shows, when electronically, the module
does not have a unique serial number. (The Crucial modules I own,
each have their own serial number, instead of some bogus value.)

The BIOS is normally very forgiving, when the SPD content is wrong.
The BIOS appears to have inherited not only the new code for reading
the SPD, but also continues to use the old "binary search for the
end of memory", which was used before memory DIMMs had an SPD. So
the BIOS may figure it out, even when the SPD is wrong.

What is discouraging, is the BIOS did not beep, when it had trouble
with the memory. What is encouraging, is there is no permanent damage
to the motherboard. There are some memory modules that ship, with
an electrical fault on the DIMM power rails, which can damage stuff.

You could try the memory in another computer, because at least Apacer
claims they use 64Mx8 chips, so that 1GB module should work in any
computer that accepts 1GB DDR. If you can manage to run the memory
in another computer, use CPUZ from, to list the contents
of the SPD chip. That might lend a clue as to why it is misbehaving.

First  |  Prev  | 
Pages: 1 2
Prev: SATA Drives - set as IDE or AHCI
Next: WOL on P5Q